Introduction

Solid oxide cells (SOCs) exhibit an efficient energy conversion in both fuel and electrolysis cells. Their capabilities and traits predetermine the SOCs for large-scale stationary applications. The effectivity of SOCs lies in the high operation temperature of 700–800 °C leading to favorable physicochemical conditions. At such temperatures, the kinetics of electrochemical reactions is significantly accelerated. In addition, the equilibrium voltage of water splitting is lower by 20% compared to conventional low-temperature processes. Although SOCs have been intensively studied for more than 30 years, there is a considerable room for improvement of the performance, which will be addressed in this study.

The SOCs consist of three components: an oxygen electrode, a fuel electrode, and an electrolyte. Electrolytes must be dense and nonconductive to electrons in order to prevent gas cross-over and parasitic currents. Electrodes need to be highly conductive for electrons, electrochemically active toward half-cell reactions, and porous to ensure the mass transport of reactants/products. Because the whole system is compact, the electrode materials must be inert to the electrolyte to prevent the formation of any undesired phases and exhibit a thermomechanical compatibility with electrolyte (i.e., thermal expansion coefficient) to prevent delamination.

The performance of the electrodes is governed mainly by three parameters: kinetics of electrochemical reactions, electrical conductivity, and transport of reactants and/or products. At the same time, contact of the three phases, triple phase boundary (TPB), is required for the electrochemical reactions to take place. In the case of electrodes exhibiting exclusively electronic conductivity, the TPB length is limited only to the contact with the electrolyte component, leading to the limitation of performance. The introduction of an electrolyte material into the electrode, thus forming a composite electrode, results into a significant increase of the TPB length. Today, composite electrodes represent the state of the art. Namely, lanthanum strontium manganite (La1-xSrxMnO3, LSM) paired with yttria-stabilized zirconia (Y2O3·ZrO2, YSZ) oxygen electrode; nickel paired with YSZ fuel electrode. However, the electrolyte phase exhibits negligible area specific electrical conductivity in comparison to both LSM and Ni [1]. This leads to a significant decrease in the total electrode conductivity of the composites compared to the single-phase electrodes [2]. Thus, the optimization of the composite electrode lies in the trade-off between kinetics, conductivity and thermomechanical compatibility.

The optimal trade-off can be achieved by grading the composition of the composite electrode, separating the electrode into two distinct layers [3]. First, an electrocatalytically active layer composed of the state-of-the-art composite material in direct contact with the electrolyte. Second, a highly conductive layer with minimal ohmic resistance composed of single-phase electrode material. This layout allows for harnessing only the positive traits of composite materials, while retaining highly conductive electrodes body.

The aim of this work is to revisit the topic of electrode composition by determining the optimal composition and structure of the electrodes for use in SOCs.

Results and discussion

In this section, the oxygen electrode half-cell will be referred to as OE, while the fuel electrode half-cell will be referred to as FE. Note that the half-cells include both the electrode and a part (ideally a half) of the electrolyte in terms of ohmic resistance. This effect is caused due to the shift of reference electrode potential to the middle of the electrolyte, as shown by McIntosh et al. [4].

The structure of the layered electrodes (consisting of three layers) will be expressed by number of layers of each material separated by colon 'pure electrode material: composite material' (LSM:LSM-YSZ, Ni:Ni-YSZ for OE and FE, respectively) as depicted in Fig. 6. Thus, 2:1 OE refers to an oxygen electrode half-cell with one layer of LSM-YSZ and two layers of pure LSM; the layer of composite material is always in a direct contact with the electrolyte.

Oxygen electrode

The area specific conductivity values of a given OE configuration at different temperatures and different gas input compositions are shown in Fig. 1. There was no significant impact of gas composition on the electrical conductivity of the OEs. The behavior of the studied configurations followed the principles stated in the previous section – 3:0 OE exhibited the maximum value of conductivity on par with 2:1 OE as one LSM-YSZ layer did not exhibit a significant resistance surplus. On the contrary, 0:3 OE exhibited a significantly lower conductivity caused by the combination of higher electrical resistance of the composite and insufficient geometrical current distribution, significantly decreasing the electrode effective cross-section.

Fig. 1
figure 1

Area specific conductivity of half-cells with given oxygen electrode configuration at different temperatures: 700 °C (green), 750 °C (blue), and 800 °C (red): A 100 + 00 cm3 min−1; B 20 + 80 cm3 min.−1 O2 + N2 (color figure online)

Full size image

Values of exchange current density j0 (calculated according to Eq. (3) with the estimated αa = αc = 0.55) of the given OE configuration at different temperatures and different gas inlet composition are shown in Fig. 2. Asymmetry of the multistep reaction mechanisms of the anodic and cathodic reactions caused non ideal conditions resulting in sum of αi is not equal to 1. The trends significantly differ for the different gas compositions. For 100 + 0 cm3 min−1 O2 + N2, the pure LSM 3:0 OE exhibited the maximum j0 (2000 A m−2 at 800 °C) at all temperatures with a gradually decreasing trend of j0 with OE composition to the lowest values determined for pure LSM-YSZ 0:3 OE. Quite different results were achieved with gas flow composition 20 + 80 cm3 min−1 O2 + N2; at 800 °C, 2:1 OE exhibited the maximum j0 (1300 A m−2) followed by 1:2 and 3:0 OE; while at other temperatures, the three OEs were on par with each other. The different results between the 100 + 0 and 20 + 80 cm3 min−1 O2 + N2 gas compositions reflected the ability of the electrodes to utilize the TPB length in their body. The kinetic performance of 3:0 OE was not impaired despite the fact that the TPB was only located at the electrode–electrolyte interface due to oxygen-rich environment and partial O2− conductivity [5]. However, in the lower oxygen concentration, the j0 values of 2:1 and 1:2 electrodes stood out due to their ability to shift the reaction zone to the electrode body, thus compensate for the effect of a lower oxygen concentration. This has a significant consequence, since in reality air is usually utilized instead of pure O2 in SOCs.

Fig. 2
figure 2

Exchange current density of given oxygen electrode configuration at different temperatures: 700 °C (green), 750 °C (blue), and 800 °C (red): A 100 + 0 cm3 min−1; B 20 + 80 cm3 min.−1 O2 + N2 (color figure online)

Full size image

Fuel electrode

The area specific conductivity values of the given FE configuration at different temperatures and different gas inlet composition are shown in Fig. 3. Inverse proportion of nickel conductivity to temperature was not captured by the study due to the strongly exponential temperature dependence of the YSZ electrolyte present in the half-cell in accordance to the principles of placement of the reference electrode [4]. The conductivity values did not show dependence on the composition of the inlet gas. At all temperatures, 1:2 and 0:3 FE configurations are on par in terms of conductivity values. Thus, the results suggest that a thin or none Ni surface layer is required for homogeneous current distribution across-the entire geometry due to the sufficiently high conductivity of the Ni-YSZ composite.

Fig. 3
figure 3

Area specific conductivity of half-cells with given hydrogen electrode configuration at different temperatures: 700 °C (green), 750 °C (blue), and 800 °C (red): A 75 + 25 cm3 min−1; B 50 + 50 cm3 min−1; C 25 + 75 cm3 min−1 H2 + H2O (color figure online)

Full size image

In contrast to OE, pure Ni 3:0 FE exhibits the lowest conductivity values. The reason for the poor 3:0 FE conductivity is the electrode delamination from the electrolyte as documented in Fig. 4A, caused by the difference in the thermal expansion coefficient between YSZ and Ni (10.5 × 10–6; 13.3 × 10–6 K−1) [6]. In contrast, Fig. 4B shows 0:3 FE with well-developed electrode–electrolyte interface. The results indicated that the YSZ phase helps to mitigate thermally induced mechanical stress.

Fig. 4
figure 4

SEM micrographs of A delaminated 3:0 Ni:Ni-YSZ configuration; B well-developed electrode–electrolyte interface of 0:3 Ni:Ni-YSZ configuration

Full size image

The values of the exchange current density (calculated according to Eq. (3) with the estimated αa = 0.65, αc = 0.50) of the given FE configurations at different temperatures and different gas inlet compositions are shown in Fig. 5. According to results, the electrode structure 1:2 FE configuration exhibited maximal values of exchange current density at all given temperatures and gas compositions. In this regard, 0:3 FE contained substantially higher TPB length, however, a significant fraction of TPB was not accessible due to a long distance from the electrolyte leading to high ionic resistance between the TPB fraction and the surface of the electrolyte.

Fig. 5
figure 5

Exchange current density of given hydrogen electrode configuration at different temperatures: 700 °C (green), 750 °C (blue), and 800 °C (red): A 75 + 25 cm3 min−1; B 50 + 50 cm3 min−1; C 25 + 75 cm3 min−1 H2 + H2O (color figure online)

Full size image

The results indicated that the highest performance can be achieved with either a Ni-YSZ electrode or a relatively thick Ni-YSZ layer covered by a thin layer of Ni. For operation at 800 °C, two layers (ca. 25 μm) of Ni-YSZ are enough to fulfill the required length of TPB; while a thicker Ni-YSZ layer (ca. 40 μm) is required for operation at temperatures below 800 °C.

Conclusion

In the case of the oxygen electrode, a thin surface layer of LSM-YSZ composite covered by LSM is required for the optimal performance, achieving both high electrical conductivity and exchange current density. However, in the case of applications with concentrated oxygen at the electrode, a pure LSM electrode should be utilized, achieving significantly higher kinetic performance due to the partial ionic conductivity of the material. The use of the pure LSM-YSZ composite should be avoided due to insufficient geometrical current distribution within the electrode body.

The fuel electrode should always contain a significant thickness of Ni-YSZ to increase the TPB length. At the same time, YSZ is a crucial component of the composite mitigating the mechanical stress at the electrode–electrolyte interface.

Experimental

Commercially available electrolyte, oxygen, and fuel electrode material powders were used in this study: yttria-stabilized zirconia (8 mol% Y2O3 •ZrO2, YSZ; TZ-8Y-SB, TOSOH); lanthanum strontium manganite ((La0.8Sr0.2)0.95MnO3-δ, LSM; LSM-P, fuelcellmaterials.com), LSM-YSZ (50:50 LSM:YSZ by wt., LSMYSZ-P, fuelcellmaterials.com); NiO (NiO-F, fuelcellmaterials.com), NiO-YSZ (50:50 Ni:YSZ by vol. after reduction, NiYSZ-P, fuelcellmaterials.com). Commercially available α-terpineol (PENTA) was mixed with electrode powders to produce electrode inks.

The electrochemical characterization of the studied electrodes was performed using in-house fabricated symmetrical cells in a three-electrode arrangement. Symmetrical cells 'oxygen electrode | electrolyte’ and ‘fuel electrode | electrolyte’ were supported by fabricated electrolyte wafers.

Electrolyte wafer fabrication

A defined amount of YSZ powder was transferred to a circular press mold, 3.6 cm in diameter, and pressed with a force of 60 kN by a hydraulic uniaxial press (Trystom s.r.o., Czechia). The resulting electrolyte green body was sintered in an elevator furnace (Clasic CZ s.r.o., Czechia) at 1360 °C for 6 h. Sintered wafers were used as substrates for electrode deposition.

Symmetrical cell fabrication

See the preparation conditions and electrode composition for oxygen and fuel electrode symmetrical cells in Table 1. First, three layers of counter (CE) and reference (RE) electrodes were deposited on the electrolyte wafer by the screen printing method, forming a body ca. 40 μm in thickness. CE formed a circle in the center of the wafer, while RE was in the form of a concentric ring with a gap between each of the electrodes following the requirements presented by Adler et al. [7]. After deposition, CE and RE were sintered.

Table 1 Conditions and materials in the preparation of oxygen electrode (Ox.) and fuel electrode (Fu.) symmetrical cells
Full size table

Consequently, three layers of working electrode (WE) were deposited and sintered in the same manner as CE and RE. Each WE layer consisted of either active material, i.e., electrode material-YSZ, or pure electrode material. The prepared oxygen WE configurations are shown in Fig. 6, the fuel WE configurations are analogous, while NiO substitutes LSM. The electrode-YSZ material is always in direct contact with the wafer working as an “active” layer, while the pure highly conductive electrode material works primarily as a geometrical current distribution layer.

Fig. 6
figure 6

Prepared three layer oxygen electrode configurations: A 3:0, B 2:1, C 1:2, D 0:3 LSM:LSM-YSZ layers

Full size image

Initialization of the electrochemical characterization

The symmetrical cells were placed in a zirconia housing equipped with golden current leads for CE and WE, platinum wiring for RE, and gas inlet/outlet capillaries. A mica plate was located in the center of the housing to separate CE + RE and WE atmospheres. The temperature of the system is accurately controlled by a thermocouple located in the vicinity of the housing. The entire setup was located in a vertical tubular furnace. The cells were heated to the initial temperature. Both types of symmetrical cells had a specific activation procedure due to the stabilization of current contacts concerning the oxygen electrode symmetrical cell; the need to reduce NiO to Ni in the fuel electrode. The electrochemical characterization of the cells was performed by Zahner Zenium PRO (Zahner-Electric, Germany) and Solartron SI 1287 + 1260 (Ametek, United Kingdom).

The oxygen electrode symmetrical cells were activated by applying a cyclic voltammetry procedure [8]: < OCP; 1; OCP; – 1 > V, 10 mV s−1, 24 cycles.

The FE cells were activated by feeding pure hydrogen into both electrode compartments while repeating the measurement of impedance spectra by electrochemical impedance spectroscopy (EIS), until stabilization, i.e., decrease of ohmic resistance to a constant value indicating a total reduction of NiO to Ni: 65 kHz to 1 Hz, open circuit potential (OCP), 50 mV amplitude of perturbing signal.

Electrochemical characterization procedure

It was crucial to ensure a constant composition of gases fed in the CE + RE compartment throughout all experiments to ensure a constant potential of the RE: oxygen electrode – 100 cm3 min−1; fuel electrode – 90 + 10 cm3 min−1 H2 + O2. Stationary i-V curves were obtained by current density \(j\) values at a certain applied potential \({E}_{\mathrm{appl}}\): OCP, 0, 0.25, 0.5, 0.75, 1.00, – 0.25, – 0.5, – 0.75, and – 1.00 V vs. RE. At each point, the stationary regime was ensured by applying the potential value for 5 min. After each value of stationary current was obtained, an impedance spectrum was recorded by EIS at the same potential. The measurements were performed at 700, 750, and 800 °C. For WE gas compositions: oxygen electrode – 100 + 0, 20 + 80 cm3 min−1 O2 + N2; fuel electrode – 75 + 25, 50 + 50, 25 + 75 cm3 min−1 H2 + H2O.

Data processing: half-cell specific conductivity and exchange current density

To evaluate the performance of the electrode configurations, two parameters were determined based on the measured data: half-cell specific conductivity σsp and exchange current density j0. Firstly, half-cell ohmic resistance values Rohm at measured potentials were determined by fitting of the impedance responses by an equivalent electronic circuit (EEC) shown in Fig. 7. The equivalent circuit consisted of the stray inductance of the measurement setup L, the desired half-cell ohmic resistance Rohm and two parallel combinations of resistance and CPE elements corresponding to low-frequency (LF) and high-frequency (HF) phenomena [9, 10]. Examples of fitted oxygen and hydrogen electrode electrochemical impedance spectra are shown in Fig. 8.

Fig. 7
figure 7

Equivalent circuit used for fitting of impedance response data

Full size image
Fig. 8
figure 8

Examples of EEC fitting of obtained impedance spectra recorded at 800 °C, OCP: A 2:1 OE; 100 + 0 cm3 min−1 O2 + N2; B 0:3 FE; 50 + 50 cm3 min−1 H2 + H2O

Full size image

Area specific half-cell conductivity was calculated based on the EIS fitting:

$${\sigma }_{\mathrm{sp}}=\frac{1}{{AR}_{\mathrm{ohm}}}$$
(1)

where \(A\) is the area of WE.

To determine j0, Emeas was corrected on the Rohm and open circuit potential EOCP:

$$\eta ={E}_{\mathrm{appl}}-{E}_{\mathrm{OCP}}-jA{R}_{\mathrm{ohm}}$$
(2)

where η was the WE overpotential. Resulting i-η data were fitted using pseudo–Butler–Volmer kinetics:

$$j={j}_{0}\left(\mathrm{exp}\left(\frac{{\alpha }_{\mathrm{a}}F\eta }{RT}\right)-\mathrm{exp}\left(-\frac{{\alpha }_{\mathrm{c}}F\eta }{RT}\right)\right)$$
(3)

where αa, αc are anodic and cathodic charge transfer coefficients respectively, F is the Faraday constant, R is the universal gas constant, and T is the thermodynamic temperature. The fitting procedure was carried out in two steps: (i) fitting of the i-η curves with all kinetic parameters j0, αa, αc in order to estimate the values of mechanistic parameters αa, αc; (ii) fitting of the i-η curves with mechanistic parameters set to a constant value to obtain representative values of j0. The use of exponential Butler–Volmer kinetics is not formally correct due to the fact that electrode reactions in SOCs are not driven by the charge-transfer reaction step [11]. However, this study took advantage of the exponential behavior of the kinetics that allowed for the estimation of a single kinetic parameter j0 facilitating the comparison between the electrode configurations [12].